Re: [Help-bash] Bash 2.05b build failure on OSX10.4

Re: [Help-bash] Bash 2.05b build failure on OSX10.4

[Geoff Down]

On Mon, Oct 13, 2014, at 02:07 PM, Chet Ramey wrote:
> On 10/11/14, 7:18 PM, Geoff Down wrote:
> 
> > configure.ldflags-append \
> >     "-Wl,-search_paths_first -lncurses"
> > configure.args          --mandir=${prefix}/share/man \
> >                         --infodir=${prefix}/share/info \
> >                         --without-installed-readline \
> >                         CFLAGS_FOR_BUILD="[get_canonical_archflags]"
> > 
> 
> > If so can you interpret them into what I should type as part of the
> > manual build process please?
> 
> Sorry, beyond noting that the -search_paths_first option looks
> interesting,
> I am not sure.  It's been a long time since I looked at 10.4.  I note
> that
> bash currently doesn't use any special linker arguments when building on
> 10.8/10.9, so that stuff probably went out of the source tree a while
> back.
> 
> Chet
> 
Thanks,
 I tried 
LDFLAGS='-search_paths_first -lncurses'
CFLAGS='-O -g -mmacosx-version-min=10.4 -arch ppc -DSSH_SOURCE_BASHRC'
./configure --without-installed-readline
 and configure ran without error or warning. But make still fails:

"expr.c:173: warning: conflicting types for built-in function 'exp2'
...
getenv.c: In function 'setenv':
getenv.c:172: warning: assignment discards qualifiers from pointer
target type
...
bind.c: In function 'rl_function_of_keyseq':
bind.c:584: warning: comparison is always true due to limited range of
data type
bind.c: In function 'rl_variable_bind':
bind.c:1394: warning: passing argument 1 of 'bool_to_int' discards
qualifiers from pointer target type
...
display.c: In function '_rl_move_cursor_relative':
display.c:1607: warning: passing argument 1 of '_rl_col_width' discards
qualifiers from pointer target type
display.c: In function 'rl_character_len':
display.c:1681: warning: comparison is always true due to limited range
of data type
...
terminal.c: In function 'get_term_capabilities':
terminal.c:353: warning: passing argument 1 of 'tgetstr' discards
qualifiers from pointer target type
...
histexpand.c: In function 'get_history_event':
histexpand.c:214: warning: passing argument 1 of '_rl_adjust_point'
discards qualifiers from pointer target type
histexpand.c:215: warning: passing argument 1 of '_rl_get_char_len'
discards qualifiers from pointer target type
...
glob.c: In function 'glob_pattern_p':
glob.c:127: warning: pointer targets in passing argument 1 of
'internal_glob_pattern_p' differ in signedness
glob.c:134: warning: pointer targets in passing argument 1 of
'internal_glob_pattern_p' differ in signedness
...
smatch.c: In function 'xstrmatch':
smatch.c:371: warning: pointer targets in passing argument 1 of
'internal_strmatch' differ in signedness
smatch.c:371: warning: pointer targets in passing argument 2 of
'internal_strmatch' differ in signedness
smatch.c:392: warning: pointer targets in passing argument 1 of
'internal_strmatch' differ in signedness
smatch.c:392: warning: pointer targets in passing argument 2 of
'internal_strmatch' differ in signedness
sm_loop.c: In function 'parse_collsym':
sm_loop.c:260: warning: pointer targets in passing argument 1 of
'collsym' differ in signedness
sm_loop.c: In function 'gmatch':
sm_loop.c:347: warning: pointer targets in passing argument 2 of
'is_cclass' differ in signedness"
 and then all those 'multiple definitions of symbol' errors I reported
 before. Is any of this stuff diagnostic?
GD

-- 
http://www.fastmail.fm - The way an email service should be